Sensors & Behavior (ARPA-E) > Data Evaluation and Modeling
Social media analytics with Twitter explorer
Understanding conversations in social media could provide program planners and communication managers unique insights into trending thoughts about energy efficiency and climate change. In 140 characters or less, the concerns, interests and public narratives about energy efficiency and climate change can be identified through tweets.
Twitter conversations reflect the social aspects of information diffusion through conventions such as retweets and other conversational responses, through the membership and social distance of these conversations, and in changes in frequency and form of the networks. This project tracked and analyzed Twitter conversations related to energy efficiency and climate change to gain insights about consumer attitudes and behavior. Our broad research question asked, “How can social media reflect consumer sentiment about energy and campaigns to reduce residential use of energy?”
This project collected Twitter data between September 2010 and March 2013, capturing more than 3 billion filtered tweets that used the terms relevant to this study. Three techniques were used to cull discernable patterns from the large quantities of data and portray the data visually: content analysis of the full Tweets, network analysis of co-occurring hashtags, and semantic analysis of the co-occurring.
Among the findings:
Conversations about energy-related issues are taking place in social media, specifically Twitter, and these communications can be studied to better understand how to use technologically-enhanced word-of-mouth to stimulate user-generated persuasion.
Using content analysis of full Tweets, network analysis of co-occurring hashtags, and semantic analysis of the co-occurring hashtags and their authors, researchers identified descriptors, concerns, actions and issues.
Researchers confirmed that studying Twitter communications can provide actionable means for assessing engagement, identifying influencers, and identifying word-of-mouth communities that can accelerate change in energy efficiency behaviors.
This study demonstrated the feasibility of using data mining techniques to gather and analyze vast amounts of data from ongoing social media conversations and of analyzing the data for meaningful metrics that describe conversations about energy consumption behavior.
The project also demonstrated tools that permit visualization of vast quantities of user-generated content about energy and sustainability.
Based on our initial results, the investigators recommend continued collection of data and development of analytical methods and tools that can: track public opinion related to energy consumption; analyze domain-specific, user generated content on social media platforms; identify and track indicators such as semantics and social roles; identify and explore patterns and disruptions; identify and benchmark grassroots resources such as author networks; characterize opportunities for resource transformation; and build semantic models to understand the aggregations of conversation streams. The Twitter Energy data is available for other researchers.
Self-organizing communities of consumers share many of the characteristics of issue publics, and further research on similarities and differences to other issue publics is needed in order to understand how to create, grow and sustain word-of-mouth persuasion for energy behavior change.